Biosketch
Maria is a PhD candidate at the Chair for Mathematical Foundations of Artificial Intelligence at LMU, the Munich Center for Machine Learning and relAI. Her research focuses on understanding the training dynamics that contribute to generalization, a key factor in ensuring the reliability of neural networks.
Maria earned her Bachelor’s in Mathematics at University of Bonn, and her Master’s degree in Mathematics in Data Science at Technical University of Munich. During her studies, she gained international exposure from academic stays in Zurich, Gothenburg and Osaka, as well as Machine Learning industry experience at Google, Fraunhofer Society and two startups. In 2024, she won the Civic Engagement award of the German Academic Scholarship Foundation (Studienstiftung) for co-founding and chairing the non profit organization www.lern-fair.de.
relAI Research
Implicit Bias and Generalization in Neural Netoworks
In my research, I investigate why artificial neural networks tend to learn general patterns instead of simply memorizing data, despite their capacity to do so. I focus on the “implicit bias” of training algorithms and study how the training process itself leads to generalization. Methodologically, I combine theoretical analyses of simplified models, enabling precise mathematical insights, with large-scale numerical experiments on realistic models, including LLMs. A deeper understanding of these internal mechanisms aims to improve the reliability and trustworthiness of AI systems and to inform the design of more robust models and learning methods.
Publications
https://arxiv.org/abs/2505.21423
https://arxiv.org/abs/2505.20076
https://arxiv.org/abs/2604.17633